Chapter Six: Conclusions and recommendations
6.1 Main findings
primarily responsible for the overall fall m the absolute number of self-employed over the period.
The results also showed the skewed composition of the self-employed by race. Not only are Whites disproportionally represented in fonnal sector self-employment but the growth in fonnal sector self- employment over the period has been driven by increases among Whites. Similarly, the results indicate that Blacks are disproportionally represented in infonnal sector self- employment and that the job losses in infonnal self-employment have occurred primarily among Blacks.
Average hourly earnmgs of the self-employed have increased over the period and this is consistent across both sectors of the economy with most of the earnings growth occurring in fonnal sector self-employment. As with previous studies that focus on the earnings of all the employed across race groups in South Africa (Hinks, 1999; Allanson, Atkins & Hinks, 2000;
Mwabu & Schultz, 2000; Erichsen & Wakeford, 2001; Rospabe, 2002), I also identify a racial hierarchy in earnings among the self-employed specifically, with Whites earning the most and Blacks earning the least. The results suggest clear divisions between fonnal and infonnal sector self-employment, where average earnings are considerably higher in fonnal sector self- employment. With Whites being over-represented in fonnal sector self-employment and with the growth in fonnal sector self-employment being driven by increases among Whites, the racial hierarchy in earnings therefore has not been eroded over the period.
Cross-sectional descriptive analysis in Chapter Four showed that there are a number of differences in the observable characteristics of the self-employed by racial group. Self-employed Whites are, on average, older and more educated. They are more likely to be married, male, and working in the fonnal sector of the economy in a skills-intensive occupation with access to fonnal credit. These characteristics are typically associated with higher returns and success in self-employment. Furthennore, the results also indicate that self-employed Whites are
distributed in activities that generate greater returns to self-employment. The distribution of self- employed Whites is more concentrated in formal sector non-agricultural activities which are not own account employment, and particularly in skills-intensive legislative/managerial, professional and associate professional occupations.
In contrast, self-employed Blacks, on average, are younger and have considerably lower levels of educational attainment than others in self-employment. They are also less likely to be married, male and to live in a metropolitan area. They are more likely to work in an unskilled occupation in the informal sector of the economy and to access only informal credit. These characteristics are typically associated with lower returns to self-employment. In addition, self-employed Blacks are distributed across activities associated with lower returns to self-employment. For instance, self-employed Blacks are concentrated particularly in non-agricultural own account self-employment in the informal sector and across semi-skilled and unskilled occupations.
In Chapter Four the descriptive statistics identified the presence of substantial average earnings differences in self-employment between Blacks and Whites in South Africa. The measures of earnings inequality in Chapter Four also describe not only the level of earnings inequality among the self-employed but also the level of inequality between self-employed Blacks and Whites. In Chapter Five, I investigated this earnings differential among the self-employed with the use of multivariate econometric techniques. The first step in the econometric analysis was to probe the determinants of earnings among the self-employed by estimating earnings equations. The estimated coefficients in these earnings regression show that educational attainment IS a significant detenninant of earnings among the self-employed: higher levels of educational attainment are associated with significantly higher returns to self-employment. Self-employed Whites generate higher returns to self-employment than Blacks for equivalent levels of educational attainment. These results are consistent with the findings of other studies in the South African labour market literature (Bhorat, 2000; Mwabu & Schultz, 2000; Rospabe, 2002;
Heintz & Posel, 2008).
The estimates reveal a premium to fonnal sector self-employment over infonnal sector self- employment and are thus consistent with predictions of dual labour market theories (Fields, 2005). The gender dummy variable reveals the presence of a male earnings premium that is greater for White males than for Black males in self-employment. The earnings of the self- employed are also positively detennined by age and residence in a metropolitan area. The results of the marital status dummies reveal that earnings are positively associated with being married (not statistically significant for self-employed Whites) or previously married and that there is no earnings premium to cohabitation (these results are broadly consistent with the findings by Casale & Posel, 2007). Furthennore, earnings are positively correlated with access to fonnal credit and this is especially evident among self-employed Whites.
I then explored the earnings differential between Blacks and Whites in self-employment using the Oaxaca-Blinder decomposition technique. I decomposed the earnings gap to detennine how much is attributable to differences in the endowments (i.e. observable characteristics) of the self- employed, and how much derives from differences in the returns to these endowments.
The large positive unadjusted earnings differential implies a "raw" earnings premium in favour of self-employed Whites of approximately 238 percent. The advantage that self-employed Whites have over self-employed Blacks in tenns of observable endowments such as educational attainment, fonnal sector employment in a metropolitan area and access to fonnal credit, is responsible for approximately 55 percent of the "raw" earnings premium. After controlling for the differences in observed endowments between the two groups of self-employed, it is evident that Whites earn, on average, approximately 97 percent more than Blacks in self-employment.
The source of this adjusted earnings differential in favour of Whites, derives from differences in returns to observable endowments (i.e. coefficients). Whites have far larger returns to educational attainment, fonnal credit and residence in a metropolitan area than Blacks in self- employment. Differences in returns to observable endowments account for approximately 45 percent of the "raw" earnings premium.
It is possible that discrimination may be responsible for these differences in the returns to observable endowments. Discrimination may take the form of consumer discrimination or discrimination in access to certain markets such as the credit and input markets. However, it is also possible that the residual earnings gap reflects the effects of omitted variables, such as unobserved characteristics, in the analysis. For example, the earnings of a self-employed individual may be affected by that individuals' business acumen, attitude toward risk, or other unmeasured labour market skills. Differing levels of social capital (or relationships of trust and reciprocity) among the self-employed may also detennine their earnings but such information is hard to measure and is therefore omitted. As a result it must be recognised that omitted variables may bias the results of the analysis.
Individuals who are self-employed may also not be a random sample of the employed, and the self-employed may be different in unobservable ways to the employed. Given the complex nature of selection in South Africa, I do not control for selection bias in my estimations.
Therefore it seems unlikely that selection would account for why Whites in self-employment earn almost double what Blacks earn, even after controlling for differences in unobservable characteristics.
6.2 Recommendationsand policy considerations
This study does not investigate explicitly why self-employment in South Africa remains so low in the face of very high rates of unemployment. The focus, rather, is on the returns to self- employment and a consideration of why average returns to self-employment among Blacks remain so much lower than among Whites. However, it must be noted that the presence of low returns to self-employment could be a factor helping to explain why self-employment is not growing more rapidly, particularly if self-employment is riskier or generates less secure income than wage employment. This suggestion warrants further investigation in a study of "reservation earnings" in South Africa.
The results from the study point toward the importance of educational attainment In the detennination of the earnings among the self-employed: higher levels of educational attainment are closely correlated with higher returns to self-employment. Further research that looks more closely at the relationship between educational attainment and returns to self-employment may shed light onto the levels and types of education necessary for success in self-employment. For example, researchers can take a closer examination of the role of technical training and skills development in the growth of self-employment and the success of the self-employed. This research may aid the formulation of policy aimed at encouraging more successful self- employment.
On a similar note, the results also suggested the importance of access to credit in the detennination of earnings among the self-employed. Access to fonnal credit affected the earnings of the self-employed favourably and thus policies aimed at providing or making funds available (i.e. credit) to entrepreneurs may assist in the successful growth of their ventures.
Further study into the role of credit in the success of the self-employed may assist policy formulation.
Finally, the nature of discrimination in self-employment warrants further focused study.
Affirmative action policies are able to tackle discrimination among the wage employed but it is more limited in its reach to the self-employed. Tackling consumer discrimination or discrimination in markets such as the credit market is not as easy to address and thus further investigation is necessary for the formulation of policies that can counter such problems.
6.3 Concluding remarks
In the context of the high levels of unemployment and poverty present in South Africa more research into entry barriers into self-employment, the skills necessary to succeed in self- employment, the ability to access credit and any other hindrances to the growth and success of the self-employed is vital. The growth of self-employment may play an important role in reducing unemployment and thus income inequality and poverty. Self-employment is an important source of gainful employment in the economy and therefore policies aimed at encouraging and assisting the self-employed are central to the welfare of those concerned.
Therefore policies aimed at removing obstacles that prevent individuals from taking advantage of economic opportunities should be encouraged.
Itis clear from this study that a large earnings gap is present among Blacks and Whites in self- employment in South Africa. Just over half of this earnings differential is attributable to differences in endowments. The remainder of this earnings gap may reflect the effects of omitted characteristics such as attitude to risk and entrepreneurial ability, or it may reflect differences in the returns to observed characteristics. Differences in returns to observed characteristics may be the result of discrimination among the self-employed, including consumer discrimination and discrimination in access to credit or product markets. Discrimination in self- employment is a possible impediment to success in self-employment and thus future research into the nature and extent of discrimination in South Africa is necessary.